The new platform has some advantages and innovations, but it also introduces some important differences. For example, data processing times are longer than in Universal Analytics, and the data model is different. In Universal Analytics, data is collected and processed based on sessions. In Google Analytics 4, data is collected and processed based on individual events.

At first glance, Google Analytics 4 may seem to have fewer reports than Universal Analytics. However, this tool introduces the possibility of doing many things that were simply impossible before. For example, you can create personalized explorations, connect Google Analytics 4 with BigQuery, and create views with segment overlap. You can also use a data-driven attribution model to analyze conversions. Overall, Google Analytics 4 is a powerful tool that offers several advantages over Universal Analytics. If you want to get the most out of your data and make data-driven decisions, you must start using Google Analytics 4 today.

Google Analytics 4: new features to exploit the value of data

Here are some of the fetures of Google Analytics 4:

• Custom Explorations: Explore reports provide the ability to analyze more specific metrics and dimensions, apply advanced segments, custom filters, and more.
• Native connection with BigQuery: Google Analytics 4 can be connected to Google BigQuery for more advanced and customized data analysis. This connection also provides long-term data retention.
• Data-driven attribution model: Google Analytics 4 uses a data-driven attribution model that determines which touchpoints are most likely to generate conversions. This allows you to attribute results to the most significant touchpoints.

To take full advantage of these new potentials, however, it is necessary to clear up common errors and misunderstandings that can lead to confusion when analyzing data.

The data are not all the same: differences and peculiarities

When migrating to a new data collection and analysis platform, it is necessary to account for the possibility that the platforms may have some differences and peculiarities.

For Google Analytics 4 in particular, it is important to be aware of certain factors that can affect the accuracy of the data. In this article, we will discuss three of these factors:

• High cardinality of dimensions, with the presence of (other) values in the tables
• Data approximation using the new HLL++ algorithm
• Presence of the value (not set) in the Landing Pages reports

We will provide more information on each of these factors and offer some solutions to help you make your reports more useful and understandable.

1. High cardinality of dimension

When creating a report in GA4, some primary or secondary dimensions may report the value (other). This is due, in particular, to the high cardinality of the dimensions.

Cardinality refers to the number of unique values that a single dimension can take on. Each dimension can have:

• A high cardinality, as in the case of user IDs, which are as numerous as the users who viewed our website in the analyzed period
• A very low cardinality, as in the case of the login status, which can take on one of the two values, logged in or not logged in

If a dimension has a high cardinality, meaning it has more than 500 unique values in a day, the table may exceed the maximum number of rows that GA4 can handle. In this case, all the excess values will be merged into the value (other), which will reduce the granularity of your data and limit your analysis options.

To overcome this, we have identified some solutions:

• Avoid using high-cardinality dimensions like User ID and Session ID in your reports
• Reduce the cardinality of your data during collection. For example, you can use a custom JavaScript on Google Tag Manager to generalize the URLs of a page
• Access your data in BigQuery. BigQuery doesn’t have the same limitations as GA4 reports
• Use a 360 account. It costs more, but you’ll get a higher report row limit, as well as higher limits for data collection, storage, and export

2. Data approximation using the new HLL++ algorithm

When generating reports in Google Analytics 4, the platform must return a large volume of data in a limited amount of time.

To reduce memory usage and expedite display time, Google Analytics 4 utilizes the HyperLogLog++ (HLL++) algorithm to estimate data. This algorithm is designed to handle the inherent impossibility of quickly and accurately measuring a large number of elements. Specifically, HLL++ counts only a portion of the data and then makes an accurate estimate of the remaining data. This approach allows Google Analytics 4 to provide reliable data quickly.

The only alternative to this processing is to use BigQuery. BigQuery uses the COUNT(DISTINCT) function to return exact aggregation data, albeit with increased memory usage and longer execution times.

3. Presence of the value (not set) in the Landing Pages reports

To identify which user journeys achieve the best performance, it is essential to analyze the Landing Pages report. However, it is common to find the value (not set) within these reports.

This may be because, in certain sessions, a page view (or screen view) event is not logged. In other words, the session is generated by events other than the page view, such as clicking or interacting with certain content.

For example, it may happen that, once the session has expired—or, by default, after 30 minutes of inactivity—the user resumes the tab left open and interacts with the page again, starting a new session beginning from a different event from a page view and then generating a Landing Page (unset) dimension.

In other cases, however, this problem can also be caused by an incorrect configuration of the tags, which allows events to be fired before the GA4 configuration tag.

Conclusions

The adoption of a new tool always requires a period of familiarization to learn how to use it to its full potential. While the new Google Analytics platform may appear sparse and less exhaustive than its predecessor at first glance, it conceals a wealth of potential that can be multiplied through the use of data exported to BigQuery.

Furthermore, with the proper technical support, it is possible to configure Google Analytics optimally, extracting the maximum possible value from every piece of data at our disposal.